When is a Survey Realistic - ?

[FishFright]

"Properly conducted" being the salient words here.

Look how few are and how wrong they are when predicting bug national events - over the last 15 years they've been more wrong right at that.

You probably didn't notice the many that are correct but the famously wrong ones were surrounded by mass media comments and the fuss makes it memorable

 

[Drago]

You probably didn't notice the many that are correct but the famously wrong ones were surrounded by mass media comments and the fuss makes it memorable

There was an article about it a while back in the Guardian, and the conclusion was that when it came to the big issues the bulk of them had indeed been wrong or misleading in more than half of examples.

 

[presta]

According to Road cc, in a recent survey, about 'half the UK population' can't afford to buy a bike for various reasons. I won't bore you with said reasons, but reading the article, the survey was taken from 2000 people across the UK from various jobs / professions.
Hadn't realised that the UK population had dwindled to about 4000 - !

Margin of Error is dependent on absolute sample size, not relative sample size, you don't have to sample the entire population to get an accurate answer. Think of it like tossing coins plucked from a skip full of them: you'll quickly discover that the probability is 50/50 after the first few dozen.

Then there's sampling bias, which professional polling companies are expert at minimising, but increasing the sample size won't necessarily reduce it.

 

[Baldy]

Of course they can't afford a bike, they're already paying off the BMW, house, TV, streaming service, computer, and their kids from their previous marriages. No salary left for toys.